Genetic identification of bivalve larvae using DNA barcoding

The generally small size and absence of clear morphological features makes species identification of meroplankton larvae very difficult or impossible. To investigate species diversity of the most abundant group – Bivalvia – in detail, samples from IsA collected bi-weekly between December 2011 and January 2013 and preserved on ethanol were analysed combining genetic and morphological methods. Individuals sorted from the samples were categorized into larval stages (D-shape veliger, transitional veliger, eyed pediveliger) and morphological measurements were taken from photomicrographs for each individual. Besides a diagram created from photomicrographs, morphological features for D-shaped larvae were further analysed. To test if genera can be distinguished on the basis of morphometric features, a multiple analysis of variance (MANOVA) was run in R (R Core Team, 2014). A model for identification of D-shaped larvae was created using a linear discriminant analysis (hinge length, shell length and width) and the data from genetic identification.

Amplification success of gene-regions suitable for genetic barcoding varies between genes and organism groups. To decide on a suitable gene region for our Arctic marine bivalve larvae, amplification of several mitochondrial (mt) genes previously used in studies on

Bivalvia (ribosomal 12S & 16S DNA, cytochrome oxidase subunit I = COI, and cytochrome b

= cytB) were tested following Plazzi & Passamonti (Plazzi and Passamonti, 2010). Only amplification of the mt 16S rDNA worked satisfactorily on DNA from crushed larvae (primer designed by Palumbi, 1994). In total 110 positive larval amplicons and 26 positive adult amplicons were obtained. After purification, Sanger sequencing at either GATC Biotech AG or Centre of Ecological and Evolutionary Synthesis (CEES) at the University of Oslo, and quality control, 74 larvae sequences were available for further analysis. Very few DNA-sequences of bivalve species found around Svalbard are registered in the GenBank database.

Additionally, as COI is the most commonly used barcoding gene, using the 16S-sequences reduced the amount of available sequences even more. Therefore, adult bivalve specimens were collected at several locations around Svalbard and identified using morphological traits followed by DNA-extraction and sequencing (mt 16 S rDNA). The material did far from cover all species recorded in Svalbard waters, but extended the number of available reference sequences. A searchable local database was created using both own adult Bivalve sequences and Bivalve sequences from GenBank (downloaded July 15^th, 2015).

The acquired DNA sequences were manually quality screened, and contigs were built from forward and reverse sequences when both were available. Sequences from the local database and the acquired larval sequences were globally aligned followed by manual optimization of the alignment. Unique sequences were blasted against database and against the local database.

When pairwise sequence identity was 99% or higher (Feng, Li and Kong, 2011) a species name was assigned. The genetic distances were evaluated applying the Kimura 2-parameter model and a neighbour-joining tree was built for verification (Tamura et al., 2013). The unique sequences identified were submitted to GenBank.

24 Box 2: Genetic barcoding

Genetic barcoding describes the process of using a standardized short sequence of DNA to identify a species. The gene sequence used as reference is usually deposited in a major gene-databank and attached to a voucher specimen of the species. Other sequences can then be matched to the sequences available at the database. The gene chosen needs enough variability to distinguish between species. It also needs to produce a robust result under repeated sequencing and amplification. It is useful for identification of e.g. larvae of species that do not have the morphological traits to be readily identified, but depends on the availability of good reference sequences in

databases. If sequences cannot be matched to existing sequences, the construction of a phylogenetic tree can help to place the specimen together with related species. This way, even if the species cannot be determined, identification can take place on a higher taxonomic level.

25 4.2.4. Feeding experiments

The trophic position of planktonic larvae of marine invertebrates is largely unknown in the Arctic system. Therefore, feeding experiments on some of the most abundant groups present during the spring bloom were conducted. These experiments were run on natural food assemblages and under natural conditions, to be able to evaluate their impact in the field.

Feeding experiments were run 4 times with a total of 7 sets à 5 replicates each for cirripede nauplii and 1 set of replicates for polychaete larvae (tab. 1). The method described here, was used for the last two runs after several set-backs and methodological adjustments with earlier trial, resulting in a total of 3 replica-sets of cirripede nauplii feeding experiments usable for further analysis. Cirripedia nauplii and water for the feeding experiments were collected at IsA (expm. 1) or BAB (expm. 2) during times of high cirripede abundance. Samples were taken with a 10 L Niskin water sampler (KC Denmark) from 15 m depth and kept in the cooling room at 4°C close to in-situ temperature over night before the experiment. The following day, the sample was filtered through a 150 µm mesh, separating larger organisms including cirripede nauplii for the experiments, and other potential grazers and the sea water solution. A t0 sample was collected as 100 ml of filtered sea water, preserved on 1% (final solution) acidic lugol with formaldehyde buffered with hexamine (final concentration of 2%) for community analysis. The rest was used as natural feeding solution for the experiments.

Cirripede nauplii were picked under a stereomicroscope, and 20 or 40 individuals (tab. 1) were kept in 50 ml GF/F filtered seawater for each replicate until the start of the experiment (up to 1 hour at 4°C). Since the aim was to distinguish the grazing-impact of cirripede nauplii on the system under conditions resembling natural conditions, consumer concentrations close to the natural abundances encountered in the field were used (paper I). At the same time, grazer concentrations needed to be low enough as not to risk food depletion towards the end of the 24 h experiment.

Table 1: Details about the feeding experiments conducted. Grey colour indicates experiments conducted but invalid because of technical problems. Black colour indicates experiments conducted and analysed.

Date Taxa Nr. of ind. Volume [ml] Duration [h] Nr. of repl. conc. Feeding solution

Cirripedia 20 1000 24 5 100 %

Cirripedia 20 1000 24 5 50 %

Controll 0 1000 24 3 100 %

Cirripedia 20 1000 25 5 100 %

Cirripedia 40 1000 25 5 100 %

Controll 0 1000 25 3 100 %

Cirripedia 20 1250 26 5 100 %

Cirripedia 40 1250 26 5 100 %

Polychaeta 20 1250 26 4 100 %

Controll 0 1250 26 3 100 %

Cirripedia 20 1250 25 5 100 %

Controll 0 1250 25 3 100 %

03.05.2012

11.05.2012

31.05.2012

06.06.2012

26

For each series of experimental runs, five controls with no animals were run in parallel to 5 experimental replicates with cirripede nauplii added. Acid washed glass bottles (Duran, 1250 ml) were filled with 1200 ml feeding solution. To start the experiment, the 50 ml GF/F filtered water containing the nauplii were added and if needed, filtered sea water added until the bottles were topped to avoid disturbing bubbles. To keep conditions for both cirripede nauplii as well as organisms in the feeding solution as close as possible to conditions in their natural environment, the filled bottles were attached to a rope in groups of 5 bottles and gently lowered into the sea water. They were incubated at 3m depth in the fjord for around 25hrs (tab. 1, hanging from a pier (fig. 5). The filled bottles were just slightly negatively buoyant and moved by waves, which prohibited organisms from sinking to the bottom of the bottles.

Temperature, salinity and light were recorded using HOBO Micro Station Logger attached to the setup. Experiments were stopped by sieving the water through a 150µm mesh to retrieve the cirripede nauplii. They were counted again and checked for their condition and preserved in ethanol. The total volume of water was measured to check for leakages. 100ml of the feeding solution both from controls and replicates (tend), was preserved on 1% lugol (final concentration), and after 24h formaldehyde buffered with hexamine was added (2% final concentration).

Figure 5: Schematics of the experimental set-up used for feeding experiments with preparative work-flow. Brown dots indicate experimental organisms.

27

Feeding-solution samples were analysed at the Institute of Oceanology, Polish Academy of Science, for community composition and abundances. The identification process followed the method from Kubiszyn et al. (in review) and is briefly outlined below. Protists were counted from 10-50 mL subsamples, which were placed in a settling chamber for 24h, using an inverted microscope with phase and interference contrasts (Nikon 120 Eclipse TE-300).

Microplankton (>20 μm) were counted from the entire chamber under 100x magnification, while nanoplankton (3–20 μm) were counted over the length of three transverse transects at 400x magnification. Up to 50 specimens of the most numerous were counted, deciding on the number of fields counted individually. Both literature and the Nordic Microalgae web base (http://nordicmicroalgae.org) was used to place organisms into the trophic groups phototroph or heterotroph (Kubiszyn et al., 2014). For individuals that could not be identified to species level, the classification “undetermined” was used.

A student t-test was used to test for differences between controls and experiments for each taxonomic group (class) and size class (10µm-spacing), assuming equal starting conditions in all bottles. Boxplots were used for visual evaluation. As multivariate test a one-way ANOSIM was chosen both on species level, taxonomic group level and for size classes to check for differences between experimental runs and controls. To visually evaluate differences of the community composition on species level between controls and experimental samples, an nMDS was run and plotted. A SIMPER analysis provided information about the species that contributed most to the differences between controls and treatments in each experimental run.

All statistical analysis were done in R with either the base package or the vegan package (Oksanen et al., 2013; R Core Team, 2014).

28

5. Summary of main findings

5.1. Timing of meroplankton, duration and contribution to the zooplankton community - Paper I & III

Paper I was a baseline study of meroplankton seasonality in Adventfjorden, in close vicinity to UNIS to be able to sample frequently year-round. Meroplankton organisms comprised a considerable proportion of the total number of zooplankton organisms over the year at our relatively shallow sample site within Adventfjorden. They dominated the zooplankton community during the productive time of the year both in number and biomass and entirely outnumbered other groups during peak occurrences in spring and summer. During the rest of the year, meroplankton occurred sparsely. The meroplankton assemblages could be divided into 5 significantly different seasonal communities. Winter and early spring communities were poorest, while spring and summer showed exceptionally high dominance of

meroplankton and most groups occurred during this time. Autumn meroplankton composition kept an intermediate position. The most numerous groups were Cirripedia in spring, and Bivalvia in spring and summer. Bryozoans were the only taxonomic group with larvae mainly found during winter, and Gastropoda larvae were encountered throughout the year. All other groups had their main occurrence in the plankton during spring through autumn, and all groups showed multiple abundance peaks. Day length and chlorophyll a were the best predictors for meroplankton composition on this coarse taxonomic resolution, followed by hydrography.

To test if seasonal meroplankton patterns found at IsA in 2012 are general features, paper III extends the study from paper I both in space and time, including 2 more fjords and years. The main goal was to test if the strong positive correlation between meroplankton occurrence and primary production found at IsA is a common feature in Svalbard fjords and if the timing of the spring bloom could be steering the timing of maximum meroplankton occurrences. The positive correlation of chl a/fluorescence and day length with abundances of most groups was confirmed. Like in the first study, Bryozoa and Gastropoda were the exceptions. Correlations between the timing of larval abundances (measured as week of the year with maximum abundance) and the peak of the bloom, (measured as week of the year with maximum chl a/fluorescence values) and start of the bloom (when chl a/fluorescense values started to increase significantly), were tested. Peak Cirripedia larval abundance was positively correlated to the start of the bloom, while peak Bivalvia larval abundance and total meroplankton were positively correlated to the peak phytoplankton bloom. For all other groups, no significant relationships were found. Including literature data, significant positive relationships between the timing of the bloom and maximum abundances were also found for Polychaeta and Echinodermata. Only at least bi-weekly sampling showed to be sufficient to catch dynamics in the meroplankton community properly, even though monthly sampling still showed some general trends.

29

Marine invertebrate larvae comprised a significant part of the zooplankton community during the productive time of the year in all locations and years, even though their contribution was lower in the deeper fjords, especially below 100 m depth. Both contribution to total

zooplankton numbers and which groups dominated varied. Bivalvia was the most abundant group in all fjords and years and Cirripedia (IsA, BAB) and Echinodermata (RiF) were the second most abundant groups in the respective fjords.

5.2. Species composition and seasonality of the dominating meroplanktonic Bivalvia larvae - Paper II

To investigate the meroplankton composition with higher taxonomic resolution, Bivalvia, as the most abundant group, was chosen for more in-depth investigation at the IsA sampling station in Adventfjorden. Four different species of bivalve larvae (Hiatella arctica, Mya truncata, Mya sp., Serripes groenlandicus) were successfully identified by DNA-barcoding (16S ribosomal DNA) and 11 new adult bivalve sequences (mt 16S rDNA) not previously available were added to the NCBI GenBank database. Through the combination of genetic barcoding and measurements of size relationships as morphological traits, a model for D-shaped larvae and a description for the identified larvae could be developed. Only the 2 Mya species could not be separated morphologically from each other. All 4 identified species occurred during the productive time of the year, and total bivalve abundance was positively correlated with chl a concentrations, but not hydrography. Reproductive seasonality and length of occurrence in the water column varied between species. Hiatella arctica and the two Mya species had similar seasonal dynamics with seemingly two spawning periods – one during spring and one later in summer. The occurrence of their D-shaped larvae coincided with 2 different peaks in bivalve larval abundance. Serripes groenlandicus appeared to have only one spawning period in early summer during a main peak in bivalve larvae. A

comparison with literature data showed that the seasonality of the bivalve larvae identified in this study (paper II) varies throughout their geographic range.

5.3. Potential trophic impact of marine invertebrate larvae on the zooplankton community - Feeding experiments with Cirripedia nauplii

Results obtained from the feeding experiments did not allow calculations of clearance rates or observe prey preferences. Differences between pre-experiment samples, controls and

experimental bottles with nauplii were nearly absent (experimental run 2) or not significant.

The student’s t-test showed no significant differences between controls and experiments for size classes or higher taxonomic groups (class) (tab. 2). Even though the results were not significant, prey-item numbers for some phytoplankton and protozoan groups and size classes

30

were slightly higher in the bottles with nauplii compared to the controls in experimental run 1.

Also, for some prey-organism groups and size classes numbers of prey-organisms were higher after the experiments compared to pre-experimental values, both in control bottles as well as experimental bottles with cirripede nauplii. Since experiments were not run in darkness, primary production continued during the experiments and removal of other potential

grazers/predators likely reduced the predation pressure on prey organisms. Besides potential flaws and set-backs in experimental set-up, a possible conclusion is that Cirripede larvae alone cannot control primary production in a late bloom scenario even during mass

occurrences as suggested earlier (Kuklinski et al., 2013). Their excretion might even facilitate microorganism growth (Seuthe, Rokkan Iversen and Narcy, 2010). The nMDS-analysis, using non-aggregated data (species resolution with size classes), indicated slight differences in species composition between controls and feeding experiments for the first experimental run, but not the second (fig. 6). Surprisingly, differences in species composition on non-aggregated data between experimental bottles and controls were still significant for all 3 experimental runs (ANOSIM, p≤0.05), even though they were not for data aggregated to higher taxonomic groups or size classes (ANOSIM). A SIMPER analysis identified Phaeocystis pouchettii (5-10 µm), and small (3-7 µm) unidentified flagellates and monoflagellates as the organisms mostly responsible for the differences between controls and treatments in all experimental runs. No further analysis was undertaken with the obtained experimental results. Using natural concentrations of both nauplii and potential prey-organisms under near-natural conditions, the results still indicate that cirripede nauplii do not exert top down control on phytoplankton during the late bloom period. During the last experimental run, several nauplii metamorphosed into the non-feeding cypris-state, which might be a reason while even less response on any taxonomic group or size class was found during experiment 2. For further experiments, parallel incubations or all incubation in the darkness and nutrients-measurements could yield better results.

Figure 6: MDS plot for community data (species) of the feeding experiments. Stress = 0.051.

31

Table 2: Results from student t-test for taxon classes and size classes between experiment and control of the experiments per experimental run.

Experiment 1 Experiment 2a Experiment 2b

Class p df Class p df Class p df

Bacillariaphyceae 0,12 4,01 Bacillariaphyceae 0,18 4 Bacillariaphyceae 0,18 4 Choanozoa 0,59 7,89 Choanozoa NA NA Choanozoa NA NA Chrysophyceae 0,31 7,36 Chrysophyceae 0,37 4 Chrysophyceae NA NA Ciliophora 0,32 4,03 Ciliophora 0,73 6,51 Ciliophora 0,36 4,64 Cryptophyta 0,38 5,39 Cryptophyta 0,37 4 Cryptophyta 0,66 5,82 Dinoflagellata 0,17 4,41 Dinoflagellata 0,17 6,3 Dinoflagellata 0,73 6,88 Flagellata 0,37 4,14 Flagellata 0,34 5,32 Flagellata 0,68 5,91 Haptophyta 0,16 7,91 Haptophyta 0,33 4,68 Haptophyta 0,94 6,7

<10µm 0,24 4,66 <10µm 0,29 4,9 <10µm 0,84 5,88 10-20µm 0,45 6,58 10-20µm 0,16 7,51 10-20µm 0,91 7,03 20-30µm 0,34 6,44 20-30µm 0,36 4,02 20-30µm 0,18 5,87

30-40µm 0,21 5,94 30-40µm 0,32 4 30-40µm 0,21 4

40-50µm 0,63 6,58 40-50µm 0,59 4,17 40-50µm 0,21 7,37 50-60µm 0,11 4,00 50-60µm 0,6 5,43 50-60µm 0,49 4,65 60-70µm 0,18 4,02 60-70µm 0,97 7,98 60-70µm 0,53 7,79

70-80µm 0,37 4,00 70-80µm 0,18 4 70-80µm 0,18 4

80-90µm 0,73 6,56 80-90µm 0,37 4 80-90µm 0,37 4

32

6. Discussion

This thesis represents one of few Arctic year-round studies of marine benthic invertebrate larval occurrence in the plankton. From one high-frequency time series throughout a whole year in Adventfjorden (paper I), it extends the study spatially and temporally with 2 further locations and a second year of sampling in two of those testing for generality of patterns found earlier (paper III) and investigates the most abundant group Bivalvia on a more detailed taxonomic level (paper II). Even though taxonomic resolution does not match some of the other available seasonal studies (Thorson, 1936; Smidt, 1979; Norden Andersen, 1984), observations over more than one year and the large area included made it possible to test if pattern in meroplankton occurrence and timing were local or general features. For at least some of the most abundant forms, larval occurrence and with that the reproductive cycle is timed to the compressed time of primary productivity in the Arctic. The advantage for

planktotrophic larvae is obvious, but also other aspect like energy input to the adult organisms for gonad maturation, conditions for settling juveniles or predation pressure could be

important factors and are discussed below. Also potential advantages and disadvantages for different reproductive strategies are discussed briefly, focusing on the most abundant groups, as well as potential influences of climate change on the meroplankton community.

6.1. Seasonality and variability in meroplankton occurrence

Within year variation in meroplankton occurrence are generally very pronounced (Thorson, 1950; Coyle, Chavtur and Pinchuk, 1996; Sewell and Jury, 2011) and reflect to a large extend seasonality in reproductive cycles of the benthic adult individuals (Mileikovsky, 1970).

Seasonal variations in abundance of meroplankton around Svalbard were strongly correlated to chl a and/or fluorescence as estimated for photosynthetically active biomass and day-length (paper I, II, III). For some groups, the timing of the bloom and the timing of main larval abundances was correlated (paper III). Even though hydrographical processes like advection and dispersion can influence larval abundances at local scales (Pedersen, Ribergaard and Simonsen, 2005), the present study showed that observed meroplankton abundances were only to a very limited extend correlated to the hydrographical parameters measured (discussed in paper I & III).

Meroplankton dynamics within the zooplankton community

Total meroplankton dynamics and seasonality following phytoplankton bloom dynamics, had strong implications for the total zooplankton composition throughout the year. Both in ice-covered fjords as well as fjords with no ice cover, a pattern with low abundances of both holoplankton and meroplankton during winter (dominance of holoplankton organisms), were followed by a faster increase in meroplankton abundances during the bloom compared to

In document Seasonality of Meroplankton in Svalbard Waters (sider 29-0)